The application of machine learning (ML) and deep learning (DL) in radiology has expanded exponentially. In recent years, an extremely large number of studies have reported about the hepatobiliary domain. Its applications range from differential diagnosis to the diagnosis of tumor invasion and prediction of treatment response and prognosis. Moreover, it has been utilized to improve the image quality of DL reconstruction. However, most clinicians are not familiar with ML and DL, and previous studies about these concepts are relatively challenging to understand. In this review article, we aimed to explain the concepts behind ML and DL and to summarize recent achievements in their use in the hepatobiliary region.
MRI is widely used in clinical practice for detecting liver diseases. Since the introduction of gadoxetic acid, MRI has become the most effective modality for the detection and characterization of focal liver lesions. According to previous meta-analyses, the area under the receiver operating characteristic curve (AUROC) was 0.97–0.99 for the diagnosis of small hepatocellular carcinoma (≥ 2 cm) by gadoxetic-acid-enhanced MRI. Moreover, the AUROC for the diagnosis of colorectal liver metastases was significantly high (0.98). Despite gadoxetic acid’s drawbacks, its clinical utility outweighs them, making it the contrast agent of choice in routine liver MRIs. Moreover, clinically, liver MRI has become more prevalent for a quantitative assessment. Liver fibrosis can be evaluated using MR elastography; whereas, hepatic steatosis and iron overload can be evaluated using proton density fat fraction, with high accuracy and reproducibility. This article reviewed the usefulness of liver MRI, which can be a comprehensive imaging modality in clinical practice.
The liver moves with respiratory motion. Respiratory motion causes image artifacts as MRI is a motion-sensitive imaging modality; thus, MRI scan speed improvement has been an important technical development target for liver MRI for years. Recent pulse sequence and image reconstruction technology advancement has realized a fast liver MRI acquisition method. Such new technologies allow us to obtain liver MRI in a shorter time, particularly, within breath-holding time. Other benefits of new the technology and the higher spatial resolution liver MRI within a given scan time are improved slice coverage and smaller pixel size. In this review, MRI pulse sequence and reconstruction technologies to accelerate scan speed for T1- and T2-weighted liver MRI will be discussed. Technologies that reduce scan time while keeping image contrast, SNR and image spatial resolution are needed for fast MRI acquisition. We will discuss the progress of MRI acquisition methods, the enabling technology, established applications, current trends, and the future outlook.
Since its first observation in the 18th century, the diffusion phenomenon has been actively studied by many researchers. Diffusion-weighted imaging (DWI) is a technique to probe the diffusion of water molecules and create a MR image with contrast based on the local diffusion properties. The DWI pixel intensity is modulated by the hindrance the diffusing water molecules experience. This hindrance is caused by structures in the tissue and reflects the state of the tissue. This characteristic makes DWI a unique and effective tool to gain more insight into the tissue’s pathophysiological condition. In the past decades, DWI has made dramatic technical progress, leading to greater acceptance in clinical practice. In the abdominal region, however, acquiring DWI with good quality is challenging because of several reasons, such as large imaging volume, respiratory and other types of motion, and difficulty in achieving homogeneous fat suppression. In this review, we discuss technical advancements from the past decades that help mitigate these problems common in abdominal imaging. We describe the use of scan acceleration techniques such as parallel imaging and compressed sensing to reduce image distortion in echo planar imaging. Then we compare techniques developed to mitigate issues due to respiratory motion, such as free-breathing, respiratory-triggering, and navigator-based approaches. Commonly used fat suppression techniques are also introduced, and their effectiveness is discussed. Additionally, the influence of the abovementioned techniques on image quality is demonstrated. Finally, we discuss the current and future clinical applications of abdominal DWI, such as whole-body DWI, simultaneous multiple-slice excitation, intravoxel incoherent motion, and the use of artificial intelligence. Abdominal DWI has the potential to develop further in the future, thanks to scan acceleration and image quality improvement driven by technological advancements. The accumulation of clinical proof will further drive clinical acceptance.
The incidence of hepatocellular carcinoma (HCC) is still on the rise in North America and Europe and is the second leading cause of cancer-related mortality. The treatment of HCC varies, with surgery and locoregional therapy (LRT) such as radiofrequency ablation and transcatheter arterial chemoembolization, and radiation therapy being the primary treatment. Currently, systemic therapy with molecular-targeted agents and immune checkpoint inhibitors (ICIs) is becoming a major treatment option for the unresectable HCC. As the HCC after LRT or systemic therapy often remains unchanged in size and shows loss of contrast effect in contrast-enhanced CT or MRI, the response evaluation criteria in solid tumors (RECIST) and World Health Organization criteria, which are usually used to evaluate the treatment response of solid tumors, are not appropriate for HCC. The modified RECIST (mRECIST) and the European Association for the Study of the Liver (EASL) criteria were developed for HCC, with a focus on viable lesions. The latest 2018 edition of the Liver Imaging Reporting and Data System (LI-RADS) also includes a section on the evaluation of treatment response. The cancer microenvironment influences the therapeutic efficacy of ICIs. Several studies have examined the utility of gadoxetic acid-enhanced MRI for predicting the pathological and molecular genetic patterns of HCC. In the future, it may be possible to stratify prognosis and predict treatment response prior to systemic therapy by using pre-treatment imaging findings.
Purpose: To compare the quality of dynamic imaging between stack-of-stars acquisition without breath-holding (DISCO-Star) and the breath-holding method (Cartesian LAVA and DISCO).
Methods: This retrospective study was conducted between October 2019 and February 2020. Two radiologists performed visual assessments of respiratory motion or pulsation artifacts, streak artifacts, liver edge sharpness, and overall image quality using a 5-point scale for two datasets: Dataset 1 (n = 107), patients with Cartesian LAVA and DISCO-Star; Dataset 2 (n = 41), patients with DISCO and DISCO-Star at different time points. Diagnosable image quality was defined as ≥ 3 points in overall image quality. Whether the scan timing of the arterial phase (AP) was appropriate was evaluated, and results between the pulse sequences were compared. In cases of inappropriate scan timing in the DISCO-Star group, retrospective reconstruction with a high frame rate (80 phases, 3 s/phase) was added.
Results: The overall image quality of Cartesian LAVA was better than that of DISCO-Star in AP. However, noninferiority was shown in the ratio of diagnosable images between Cartesian LAVA and DISCO-Star in AP. There was no significant difference in the ratio of appropriate scan timing between DISCO-Star and Cartesian LAVA; however, the ratio of appropriate scan timing in DISCO-Star with high frame rate reconstruction was significantly higher than that in Cartesian LAVA in both readers. Overall image quality scores between DISCO and DISCO-Star were not significantly different in AP. There was no significant difference in the ratio of appropriate scan timing between DISCO-Star with high frame rate reconstruction and DISCO in both readers.
Conclusion: The use of DISCO-Star with high frame rate reconstruction is a good solution to obtain appropriate AP scan timing compared with Cartesian LAVA. DISCO-Star showed equivalent image quality in all phases and in the ratio of appropriate AP scan timing compared with DISCO.
Purpose: To assess the effect of an ultrahigh b value of 3000 s/mm2 and the minimal TE of 53 ms on image quality and T2 shine-through effect in liver diffusion-weighted imaging (DWI) using a 3-Tesla MRI scanner with a peak gradient of 100 mT/m.
Methods: At b values of 1000 and 3000 s/mm2 and at the minimal (44–53 ms) and routine TEs (70 ms), DWI of our original phantom and liver DWI in 10 healthy volunteers and 26 patients with 35 hepatic hemangiomas were acquired with this scanner, and the quantified SNR of the phantom and the hepatic parenchyma in the volunteers and the contrast-to-noise ratio (CNR) of the hepatic hemangiomas were calculated; two independent readers qualitatively graded the overall image quality in the volunteers and determined the presence or absence of the T2 shine-through effect related to the hemangiomas in the patients. We compared the SNR and subjective overall image quality between the minimal and routine TEs and the CNR and incidence of the T2 shine-through effect between b values of 1000 and 3000 s/mm2. Inter-reader agreement was also evaluated.
Results: The SNR at both b values was significantly higher, and the subjective overall image quality at a b value of 3000 s/mm2 was significantly better at the minimal TE than at the routine TE (P < 0.05 for all). The CNR at both TEs and the incidence of the T2 shine-through effect at the minimal TE were significantly lower at a b value of 3000 s/mm2 than at a b value of 1000 s/mm2 (P < 0.05 for all). Inter-reader agreement was excellent.
Conclusion: Liver DWI at the ultrahigh b value can reduce the T2 shine-through effect with improvement of image quality using the minimal TE.
Purpose: The wavelet denoising with geometry factor weighting (g-denoising) method can reduce the image noise by adapting to spatially varying noise levels induced by parallel imaging. The aim of this study was to investigate the clinical applicability of g-denoising on hepatobiliary-phase (HBP) images with gadoxetic acid.
Methods: We subjected 53 patients suspected of harboring hepatic neoplastic lesions to gadoxetic acid-enhanced HBP imaging with and without g-denoising (g+HBP and g–HBP). The matrix size was reduced for g+HBP images to avoid prolonging the scanning time. Two radiologists calculated the SNR, the portal vein-, and paraspinal muscle contrast-to-noise ratio (CNR) relative to the hepatic parenchyma (liver-to-portal vein- and liver-to-muscle CNR). Two other radiologists independently graded the sharpness of the liver edge, the visibility of intrahepatic vessels, the image noise, the homogeneity of liver parenchyma, and the overall image quality using a 5-point scale. Differences between g–HBP and g+HBP images were determined with the two-sided Wilcoxon signed-rank test.
Results: The liver-to-portal- and liver-to-muscle CNR and the SNR were significantly higher on g+HBP- than g–HBP images (P < 0.01), as was the qualitative score for the image noise, homogeneity of liver parenchyma, and overall image quality (P < 0.01). Although there were no significant differences in the scores for the sharpness of the liver edge or the score assigned for the visibility of intrahepatic vessels (P = 0.05, 0.43), with g+HBP the score was lower in three patients for the sharpness of the liver edge and in six patients for the visibility of intrahepatic vessels.
Conclusion: At gadoxetic acid-enhanced HBP imaging, g-denoising yielded a better image quality than conventional HBP imaging although the anatomic details may be degraded.
Purpose: This multi-scanner study aimed to investigate the validity of single breath-hold (BH) diffusion-weighted imaging (DWI) using simultaneous-multislice (SMS) echo-planar imaging in multiple abdominal organs to enable faster acquisition and reliable quantification of apparent diffusion coefficient (ADC).
Methods: SNR, geometric distortion (GD), and ADC in a phantom; the ADC in the liver, renal cortex, paraspinal muscle, spleen, and pancreas; and the signal intensity ratio of the portal vein-to-muscle (SIRPV-M) in healthy volunteers were compared between BH- and respiratory-triggered (RT) DWI with b-values of 0 and 800 s/mm2 in two different MRI scanners.
Results: The phantom study showed that the SNR of BH-DWI was significantly lower than that of the RT-DWI (P < 0.05 for both scanners), whereas the GD and ADC of BH-DWI did not differ significantly from those of the RT-DWI (P = 0.09–0.60). In the volunteer study, the scan times were 23 seconds for BH-DWI and 184±33 seconds for RT-DWI, respectively. The ADC of the liver in BH-DWI was significantly lower than that in RT-DWI (P < 0.05 for both scanners), whereas there were no significant differences in the ADCs of the renal cortex, paraspinal muscle, spleen, or pancreas between BH-DWI and RT-DWI (P = 0.07–0.86). The SIRPV-M in BH-DWI was significantly smaller than in RT-DWI (P < 0.05 for both scanners).
Conclusion: The proposed method enables the acquisition of abdominal diffusion-weighted images in a single BH.
Purpose: The Multi-echo Dixon (ME-Dixon) is a non-invasive quantitative MRI technique to diagnose non-alcoholic fatty liver disease (NAFLD). In this study, the hydrogen proton MR spectroscopy (1H-MRS) was used as a reference to explore the accuracy of the ME-Dixon technique in evaluating hepatic steatosis in NAFLD patients after ingesting formulated food and its correlation with changes in clinical indicators.
Methods: Twenty-seven patients with NAFLD were enrolled. Fifteen patients completed 12 weeks of treatment with prebiotics and dietary fiber. In addition, abdominal MRI scans and blood tests were performed before and after treatment. The MRI-proton density fat fraction (MRI-PDFF) and MRS-PDFF were measured using the ME-Dixon and 1H-MRS techniques. The Bland–Altman method and Pearson correlation analysis were used to test the consistency of the two techniques for measuring the liver fat content and the changed values. Besides, correlation analysis was conducted between the MRI-PDFF value and metabolic indicators.
Results: In the PDFF quantification of 42 person-times and the monitoring of the PDFF change in 15 patients under treatment, there was a good consistency and a correlation between MRI and MRS. At baseline, MRI-PDFF was positively correlated with insulin resistance index (HOMA-IR), fatty liver index (FLI), and liver enzymes. After treatment, the changes in MRI-PDFF were positively correlated with the recovery degree of FLI and liver enzymes.
Conclusion: ME-Dixon has a good consistency and a correlation with MRS in quantifying the liver fat content and monitoring the treatment effect, which may be used as an accurate indicator for clinical monitoring of changes in the liver fat content.